IP Library › Granted Patent US 12,541,777
Granted Patent B2
US 12,541,777 · App. 18/360,307 · Granted Feb 3, 2026

Counting and extracting opinions in product reviews

Inventors: Christopher Malon (Fort Lee, NJ); Hideo Kobayashi (Dallas, TX)
Assignee: NEC Corporation
G06Q30/0282G06F40/295G06F40/30
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,541,777
App. No.
18/360,307
Granted
Feb 3, 2026
Kind
B2
Abstract

A computer-implemented method for counting and extracting opinions in product reviews is provided. The method includes inputting a hypothesis opinion, a product name, and product reviews relating to a product, applying a decontextualization component to the product reviews by using the product name, applying the decontextualization component to the hypothesis opinion by using the product name, applying an entailment model to classify each sentence of the decontextualized product reviews against the decontextualized hypothesis opinion, and outputting one or more sentences classified as entailing the hypothesis opinion and a count of corresponding reviews.

Claims (40)

1 . A computer-implemented method for counting and extracting opinions in product reviews, the method comprising:

inputting a hypothesis opinion, a product name, and product reviews relating to a product;

identifying and classifying noun phrases in product reviews by using a mention type classifier, wherein the noun phrases referring to the product being reviewed are recognized using a feedforward neural network using span features from a trained transformer encoder;

applying a decontextualization component to the product reviews by using the product name;

applying the decontextualization component to the hypothesis opinion by using the product name;

excluding noun phrases referring to competing products;

applying an entailment model to classify each sentence of the decontextualized product reviews against the decontextualized hypothesis opinion as entailing the hypothesis to minimize coreference and bridging errors, the entailment model extracts sentences from product reviews that the mention type classifier predicts as entailing the hypothesis opinion and counts a corresponding number of the product reviews; and

outputting one or more sentences classified as entailing the hypothesis opinion and a count of corresponding reviews.

2 . The computer-implemented method of claim 1 , wherein product review sentences with noun phrases referring to a competing product or part/attribute of a competing product are excluded.

3 . The computer-implemented method of claim 1 , wherein product review sentences with noun phrases referring to the product being reviewed have those phrases replaced with a string indicated as “the main product”.

4 . The computer-implemented method of claim 1 , wherein product review sentences with noun phrases referring to part/attribute being reviewed have a phrase indicated as “of the main product” inserted after those sentences.

5 . The computer-implemented method of claim 1 , wherein a sentence “I bought [product name]” is inserted before the span features are computed.

6 . The computer-implemented method of claim 1 , wherein a feedforward neural network is trained to recognize noun phrase classes including at least the “main product,” “part/attribute of the main product,” a “competing product,” “part/attribute of a competing product,” and “others”.

7 . A computer program product for counting and extracting opinions in product reviews, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to perform a method comprising:

input a hypothesis opinion, a product name, and product reviews relating to a product;

identify and classify noun phrases in the product reviews by using a mention type classifier, wherein the noun phrases referring to the product being reviewed are recognized using a feedforward neural network using span features from a trained transformer encoder;

apply a decontextualization component to the product reviews by using the product name;

apply the decontextualization component to the hypothesis opinion by using the product name;

exclude noun phrases referring to competing products;

apply an entailment model to classify each sentence of the decontextualized product reviews against the decontextualized hypothesis opinion as entailing the hypothesis to minimize coreference and bridging errors, the entailment model extracts sentences from product reviews that the mention type classifier predicts as entailing the hypothesis opinion and counts a corresponding number of the product reviews; and

output one or more sentences classified as entailing the hypothesis opinion and a count of corresponding reviews.

8 . The computer program product of claim 7 , wherein product review sentences with noun phrases referring to a competing product or part/attribute of a competing product are excluded.

9 . The computer program product of claim 7 , wherein product review sentences with noun phrases referring to the product being reviewed have those phrases replaced with a string indicated as “the main product”.

10 . The computer program product of claim 7 , wherein product review sentences with noun phrases referring to part/attribute being reviewed have a phrase indicated as “of the main product” inserted after those sentences.

11 . The computer program product of claim 7 , wherein a sentence “I bought [product name]” is inserted before the span features are computed.

12 . The computer program product of claim 7 , wherein a feedforward neural network is trained to recognize noun phrase classes including at least the “main product,” “part/attribute of the main product,” a “competing product,” “part/attribute of a competing product,” and “others”.

13 . A computer processing system for counting and extracting opinions in product reviews, comprising:

a memory device for storing program code; and

a processor device, operatively coupled to the memory device, for running the program code to:

input a hypothesis opinion, a product name, and product reviews relating to a product;

identify and classify noun phrases in the product reviews by using a mention type classifier, wherein the noun phrases referring to the product being reviewed are recognized using a feedforward neural network using span features from a trained transformer encoder;

apply a decontextualization component to the product reviews by using the product name;

apply the decontextualization component to the hypothesis opinion by using the product name;

exclude noun phrases referring to competing products;

apply an entailment model to classify each sentence of the decontextualized product reviews against the decontextualized hypothesis opinion as entailing the hypothesis to minimize coreference and bridging errors, the entailment model extracts sentences from product reviews that the mention type classifier predicts as entailing the hypothesis opinion and counts a corresponding number of the product reviews; and

output one or more sentences classified as entailing the hypothesis opinion and a count of corresponding reviews.

14 . The computer processing system of claim 13 , wherein product review sentences with noun phrases referring to a competing product or part/attribute of a competing product are excluded.

15 . The computer processing system of claim 13 , wherein product review sentences with noun phrases referring to the product being reviewed have those phrases replaced with a string indicated as “the main product”.

16 . The computer processing system of claim 13 , wherein product review sentences with noun phrases referring to part/attribute being reviewed have a phrase indicated as “of the main product” inserted after those sentences.

17 . The computer processing system of claim 13 , wherein a sentence “I bought [product name]” is inserted before the span features are computed.

Assignments (2)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Dec 12, 2025
From: NEC LABORATORIES AMERICA, INC.
To: NEC CORPORATION
Reel/Frame 073203/0407 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Jul 27, 2023
From: MALON, CHRISTOPHER; KOBAYASHI, HIDEO
To: NEC LABORATORIES AMERICA, INC.
Reel/Frame 064405/0234 →
Continuity (2)
Provisional Application 63398640 · Aug 17, 2022
Related Publication 20240062256A1 · Feb 22, 2024
References Cited (34)
US 20050091038A1 · Yi · 2005 [cited by examiner]
US 20110113027A1 · Shen · 2011 [cited by examiner]
US 20150379090A1 · Gou · 2015 [cited by examiner]
US 20160180437A1 · Boston · 2016 [cited by examiner]
US 20160180438A1 · Boston · 2016 [cited by examiner]
US 20180260860A1 · Devanathan · 2018 [cited by examiner]
US 20230214888A1 · Renard · 2023 [cited by examiner]
Aravindan, S. and A. Ekbal, “Feature Extraction and Opinion Mining in Online Product Reviews,” 2014, International Conference on Information Technology, Bhubaneswar, India, pp. 94-99, (Year: 2014). [cited by examiner]
Laban, P., Schnabel, T., Bennett, P. N., & Hearst, M. A. (Feb. 9, 2022). SummaC: Re-visiting NLI-based models for inconsistency detection in summarization. Transactions of the Association for Computational Linguistics, … [cited by applicant]
Pontiki, M., Galanis, D., Papageorgiou, H., Androutsopoulos, I., Manandhar, S., AL-Smadi, M., . . . & Eryiit, G. (Jun. 16, 2016). Semeval-2016 task 5: Aspect based sentiment analysis. In ProWorkshop on Semantic Evaluati… [cited by applicant]
Pontiki, M., Galanis, D., Pavlopoulos, J., Papageorgious, H., Androutsopoulos, I., Manandhar, S. (Aug. 23, 2014) Semeval-2014 Task 4: Aspect based sentiment analysis. In Proceedings of the 8th International Workshop on … [cited by applicant]
Aone, C., & William, S. (Jun. 1, 1995). Evaluating automated and manual acquisition of anaphora resolution strategies. In 33rd Annual Meeting of the Association for Computational Linguistics (pp. 122-129). [cited by applicant]
Williams, A., Nangia, N., & Bowman, S. R. (Apr. 18, 2017). A broad-coverage challenge corpus for sentence understanding through inference. arXiv preprint arXiv:1704.05426. [cited by applicant]
Recasens, M., De Marneffe, M. C., & Potts, C. (Jun. 9, 2013). The life and death of discourse entities: Identifying singleton mentions. In Proceedings of the 2013 conference of the North American chapter of the associat… [cited by applicant]
Lee, H., Chang, A., Peirsman, Y., Chambers, N., Surdeanu, M., & Jurafsky, D. (Nov. 20, 2012). Deterministic coreference resolution based on entity-centric, precision-ranked rules. Computational linguistics, 39(4), 885-9… [cited by applicant]
Hou, Y., Markert, K., & Strube, M. (Oct. 25, 2014). A rule-based system for unrestricted bridging resolution: Recognizing bridging anaphora and finding links to antecedents. In Proceedings of the 2014 Conference on Empi… [cited by applicant]
Clark, H. H. (1975). Bridging. In Theoretical issues in natural language processing. Stanford University Journal. (pp. 1-6). [cited by applicant]
Chu, E., & Liu, P. (May 24, 2019). Meansum: A neural model for unsupervised multi-document abstractive summarization. In International Conference on Machine Learning (pp. 1223-1232). PMLR. [cited by applicant]
Kim Amplayo, R., Brazinskas, A., Suhara, Y., Wang, X., & Liu, B. (Jul. 6, 2022). Beyond opinion mining: Summarizing opinions of customer reviews. In Proceedings of the 45th International ACM SIGIR Conference on Research… [cited by applicant]
Wan, D., & Bansal, M. (May 16, 2022). FactPEGASUS: Factuality-aware pre-training and fine-tuning for abstractive summarization. arXiv preprint arXiv:2205.07830. [cited by applicant]
Schuster, T., Fisch, A., & Barzilay, R. (Jun. 6, 2021). Get your vitamin C! robust fact verification with contrastive evidence. arXiv preprint arXiv:2103.08541. [cited by applicant]
Scialom, T., Dray, P. A., Gallinari, P., Lamprier, S., Piwowarski, B., Staiano, J., & Wang, A. (Nov. 7, 2021). Questeval: Summarization asks for fact-based evaluation. arXiv preprint arXiv:2103.12693. [cited by applicant]
Cao, M., Dong, Y., Wu, J., & Cheung, J. C. K. (Nov. 16, 2020). Factual error correction for abstractive summarization models. arXiv preprint arXiv:2010.08712. [cited by applicant]
De Clercq, O., & Hoste, V. (Dec. 12, 2020). It's absolutely divine! Can fine-grained sentiment analysis benefit from coreference resolution ?. In CRAC workshop (pp. 11-21). Association for Computational Linguistics (ACL… [cited by applicant]
Suhara, Y., Wang, X., Angelidis, S., & Tan, W. C. (Jul. 5, 2020). OpinionDigest: A simple framework for opinion summarization. arXiv preprint arXiv:2005.01901. [cited by applicant]
Angelidis, S., Amplayo, R. K., Suhara, Y., Wang, X., & Lapata, M. (Dec. 8, 2020). Extractive opinion summarization in quantized transformer spaces. Transactions of the Association for Computational Linguistics, 9, 277-2… [cited by applicant]
Xu, L., & Choi, J. D. (Sep. 28, 2020). Revealing the myth of higher-order inference in coreference resolution. arXiv preprint arXiv:2009.12013. [cited by applicant]
Wang, H., Liu, B., Li, C., Yang, Y., & Li, T. (Aug. 31, 2019). Learning with noisy labels for sentence-level sentiment classification. arXiv preprint arXiv:1909.00124. [cited by applicant]
Joshi, M., Chen, D., Liu, Y., Weld, D. S., Zettlemoyer, L., & Levy, O. (Jan. 18, 2020). Spanbert: Improving pre-training by representing and predicting spans. Transactions of the association for computational linguistic… [cited by applicant]
Loshchilov, I., & Hutter, F. (Jan. 4, 2019). Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101. [cited by applicant]
Lee, K., He, L., Lewis, M., & Zettlemoyer, L. (Dec. 15, 2017). End-to-end neural coreference resolution. arXiv preprint arXiv:1707.07045. [cited by applicant]
He, R., & McAuley, J. (Feb. 6, 2016). Ups and downs: Modeling the visual evolution of fashion trends with one-class collaborative filtering. In proceedings of the 25th international conference on world wide web (pp. 507… [cited by applicant]
Mcauley, J., Targett, C., Shi, Q., & Van Den Hengel, A. (Jun. 17, 2015). Image-based recommendations on styles and substitutes. In Proceedings of the 38th international ACM SIGIR conference on research and development i… [cited by applicant]
Bahdanau, D., Cho, K., & Bengio, Y. (May 19, 2016). Neural machine translation by jointly learning to align and translate. arXiv preprint arXiv:1409.0473. [cited by applicant]